Abstract 191P
Background
This study aims to evaluate the impact of perifissural nodules (PFNs) on the false positive rate in lung cancer screening at the participant level. PFNs, recognized as benign in lung cancer screening trials, constitute a significant proportion of all nodules (20-30%), potentially influencing the false positive rate and leading to unnecessary follow-ups. This issue is particularly pertinent when AI systems serve as the primary readers, as they currently face challenges in accurately classifying PFNs.
Methods
We analyzed 1,253 baseline scans from the UK Lung Cancer Screening Trial, focusing on pulmonary nodules exceeding a volume of 15 mm³. Utilizing an AI-based software, we automatically detected and volumetrically quantified solid pulmonary nodules. Subsequently, an experienced reader visually classified all AI-detected pulmonary nodules with a volume of ≥30 mm³, distinguishing between PFNs and non-PFNs. Pulmonary nodules measuring <100 mm³ were considered negative, while those ≥100 mm³ were categorized as positive.
Results
At the nodule level, 375 pulmonary nodules were identified as PFNs, with 296 (78.9%) measuring <100 mm³ and 79 (21.1%) measuring ≥100 mm³. At the participant level, among 1,253 participants, 316 (25.2%) were found to have PFNs. Out of these, 250 (20.0%) participants had only negative PFNs, while 66 (5.2%) participants had positive PFNs. Notably, 33 (2.6%) participants with positive PFNs did not exhibit concurrent pulmonary nodules measuring ≥100 mm³.
Conclusions
The use of AI-based software as the primary reader in lung cancer screening results in a limited number of false positive PFNs.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.